[FLINK-9421] RunningJobsRegistry entries are not cleaned up after job termination

• Type: Bug
• Status: Closed
• Priority: Critical
• Resolution: Fixed
• Affects Version/s: 1.5.0, 1.6.0
• Fix Version/s: 1.5.0
• Component/s: Runtime / Coordination
• Assignee: Till Rohrmann
• Reporter: Till Rohrmann
• Created: 23/May/18 10:04
• Updated: 24/May/18 14:53
• Resolved: 24/May/18 14:53

Description

Currently, the Dispatcher does not clean up the RunningJobsRegistry after the job has finished. The consequence is that a ZNode with the JobID and a state num per job remains in ZooKeeper.

We should clean up these ZNodes to avoid a resource leak.

Issue Links

• links to: GitHub Pull Request #6068